Rapid and Economical Method for Biochemical Screening of Stool Isolates for Salmonella and Shigella Species

Abstract

The LOUIS test is a screening protocol for Salmonella and Shigella that uses rapid enzyme tests. It had a sensitivity of 100% and a specificity of 94% and achieved presumptive reporting of Salmonella 3 h after colony isolation with savings of time and money compared to commercial identification systems.

A number of different types of selective media have been developed to enhance the recovery of enteric pathogens from stool samples. These conventional media employ simple biochemical reactions, such as the nonfermentation of lactose or production of hydrogen sulfide, to discriminate between pathogenic and nonpathogenic organisms. Unfortunately, these media have been shown to have poor specificity and produce numerous false-positive colonies (3, 5, 7, 8, 11). Confirmation of suspect colonies is highly nonspecific, and this places a heavy burden on the laboratory in terms of biochemical and serological tests (3, 5, 10).

Recently several agar media have been developed that contain chromogenic substrates for enzymes specific for Salmonella. Chromogenic media have been shown to have higher specificities but lower sensitivities than conventional media(5, 7, 9). So far there have been no reports of chromogenic media suitable for the isolation of Shigella; therefore, conventional media have to be employed for this purpose.

Biochemical testing utilizes the enzymatic characteristics of microorganisms. Growth-dependent tests mostly require at least an overnight incubation period, whereas tests that use constitutively present enzymes have incubation times of seconds to 4 h (1). These rapid enzyme tests usually require heavy suspensions of fresh cultures (6). With use of microvolumes, fewer colonies are required to achieve the necessary density of suspension, and hence, they can be tested directly from primary isolation media.

This study developed a rapid, cost-effective screening test that reduced unnecessary identifications by using the minimum number of biochemical characters (4). A protocol was established based on four rapid biochemical tests: lysine decarboxylase (LDC), o-nitrophenyl-beta-d-galactopyranoside (ONPG), urea, and indole and also some serological screening (which I have named the LOUIS test).

The LOUIS test method is comprised of a two-step algorithm (Table 1). Suspect oxidase-negative colonies were inoculated to 1 ml of sterile saline to give a suspension equivalent to at least McFarland no. 4. Resulting suspensions were then dispensed (0.2 ml) into three test tubes. A 0.2-ml volume was also dispensed into a nutrient agar slant (Oxoid, Basingstoke, United Kingdom) and a cystine-lactose-electrolyte-deficient (Oxoid) purity plate. Reagent tablets (AS Rosco, Taastrup, Denmark) were then added to the tubes, and three drops of mineral oil was added to the LDC-indole double tablet and urea tubes. All were incubated at 37°C for 3 h. After scoring of the LDC reaction, three drops of Kovacs' reagent was added to that tube to detect the indole reaction. Results were compared to data in Table 1 for presumptive identification. Biochemical reactions were controlled with Salmonella enterica serovar Typhimurium ATCC 14028, Proteus mirabilis ATCC 14153, and Escherichia coli ATCC 10536.

TABLE 1.

LOUIS test algorithm

Presumptive IDb	Biochemical test result				Step 1	Step 2
	LDC	ONPG	Urea	Indole
E. coli	+	+	−	+	Discard
	+	−	−	+
Proteus spp. or Morganella morganii	−	−	+	+	Discard
	−	−	+	−
Salmonella	+	−	−	−	Confirm by serology (polyvalent and group specific)	If serology positive, identifya; if negative, discard
Shigella spp. (possible Salmonella serovar Paratyphi A)	−	−	−	−	Full biochemical identificationa	Confirm with serology; if negative, discard
Shigella spp.	−	−	−	+	Full biochemical identificationa	Confirm with serology; if negative, discard
Shigella sonnei or Shigella dsyen- teriae 1	−	+	−	−	Confirm by serology	If serology positive, identifya; if negative, discard

^aIdentification was by API rapid 20E.

^bID, identification.

Polyvalent and group-specific sera (Murex, Dartford, United Kingdom) were used to obtain serological confirmation by slide agglutination of material from the nutrient agar slant. Cystine-lactose-electrolyte-deficient purity plates were incubated overnight at 37°C. If mixed cultures were found, the organisms were retested. All clinical isolates were fully identified by using the API rapid 20E gallery (bioMerieux, Marcy l'Etoile, France) for purposes of comparison. The API rapid 20E identifications were carried out following the manufacturer's instructions and incubated for 4 h.

During a 6-month period, 2,500 stool samples were examined for production of lactose-nonfermenting colonies on MacConkey (Mac) and deoxycholate-lactose-sucrose (DCLS) primary isolation agars and either black or translucent colonies on xylose-lysine-deoxycholate agar (XLD) after enrichment in selenite F broth (all from Oxoid). This gave rise to 265 suitable isolates from primary isolation and 40 after enrichment for biochemical testing. In addition, 121 reference Salmonella strains and 21 Shigella strains were tested.

The LOUIS test algorithm correctly identified all of the stock organism collections (Table 2). The Salmonella profile correctly suggested 118 of 121 isolates showing sensitivity of 97.5%. Three Salmonella isolates were LDC negative (including the serotype S. enterica Paratyphi A). All 14 Shigella sonnei strains had a 100% correlation with the algorithm profile. The remaining Shigella spp. were split between the two Shigella profiles.

TABLE 2.

LOUIS test profiles of reference strains and clinical isolates

LOUIS profile	Reference strain(s) (na)	Isolates from Mac and DCLS (n)	Isolates from XLD (n)
+−−−	Serotype Typhi (1)d	Salmonella (49)	Salmonella (18)
	Salmonella group B (30)	Hafnia alvei (3)
	Salmonella group C (17)
	Salmonella group D (68)
	Salmonella group E (2)
++−+ or +−−+		Escherichia coli (119)
−−+− or −−++		Proteusb (53)	Proteusc (2)
−−−−	Serotype Paratyphi A (1)	Salmonella (1)	Proteusb (18)
	Salmonella group C (2)	Serratia spp. (1)
	Shigella flexneri (3)	Enterobacter spp. (1)
	Shigella boydii (1)	Citrobacter spp. (2)
−−−+	Shigella flexneri (1)	Shigella flexneri (1)
	Shigella boydii (2)	Citrobacter spp. (2)
		Escherichia coli (4)
−+−−	Shigella sonnei (14)	Citrobacter spp. (22)	Citrobacter spp. (2)
		Enterobacter spp. (7)

^an, no. of strains.

^bProteus group includes P. mirabilis, P. vulgaris, and Morganella morganii.

^cTwo isolates of M. morganii.

^dSalmonella enterica serotype Typhi.

Step 1 of the LOUIS test when applied to the 265 clinical isolates from Mac and DCLS (Table 2) allowed 76.9% of the isolates to be discarded as nonsignificant. Of the 52 presumptive Salmonella profiles, 49 had positive polyvalent and group-specific serology. The remaining three were identified as ONPG negative Hafnia alvei. The algorithm for the screening for Salmonella showed sensitivity of 98% (95% confidence interval [CI], 89.5 to 99.6) and specificity of 98.6% (95% CI, 96 to 99.5). The application of step 2 of the algorithm resulted in 12 unknown organisms that required identification. Of these, only two were significant isolates; there was one LDC-negative Salmonella isolate and one indole-positive Shigella flexneri isolate. Overall the LOUIS test showed sensitivity of 100% (95% CI, 93 to 100) and specificity of 94% (95% CI, 89.9 to 96.4).

When the LOUIS test was applied to isolates from XLD agar, the urea test did not produce positive results for the Proteus strains after the 3-h incubation period (Table 2). This anomaly was confirmed by AS Rosco: growth in XLD medium results in delayed positives for urease production for most Proteus strains. Therefore, use of the LOUIS test with XLD is not recommended unless a urea-tryptophan deaminase double tablet is used instead of urea alone. A positive tryptophan deaminase reaction is more specific for Proteus and Providencia spp. and is not affected by growth on XLD.

The LOUIS test is cost effective compared to the use of commercial identification systems. The cost is £0.97 (£1 = US$1.81 in 2004 British pounds) per isolate compared to £2.98 for the API system employed, representing a 67% saving. The use of commercially available reagent tablets makes this test easy to perform, and the use of microvolumes of bacterial suspensions requires only a few colonies to enable testing (two to four colonies). Apart from the reagents, all other necessary components of the test are readily available in the laboratory. The protocol is simple, straightforward, and fits easily into the routine workload of the laboratory. In comparison to a previously reported rapid screening system (2), only one set of tests is inoculated in order to differentiate between organisms.

Presumptive reporting of Salmonella and S. sonnei can be achieved after 3 h, and final reporting times of common enteric pathogens can be significantly reduced. This could be useful in outbreak situations, both nosocomial and community based, where large amounts of screening can be involved. It has been observed in hospital outbreaks that by the time Salmonella has been detected it had already spread to other patients (12, 13).

This study is limited in that it has been evaluated for use only with three medium types, and for one of those it is not recommended. Further testing on other types of medium is recommended.